S24A-01
Subduction Zone Geometry and Pre-seismic Tectonic Constraints From the Andaman Micro- plate Region.
The 2004 Sumatra–Andaman mega-thrust rupture broke along the narrow fore-arc sliver boundary of the Indo- Burmese collision. Earlier events of 1679 (M~7.5), 1941 (M 7.7), 1881 (M~7.9) and 2002 (Mw 7.3) generated spatially restricted ruptures along this margin. Spatio-temporal analysis of the pre-seismic earthquakes showed dense seismicity in the back-arc region but negligible activity towards the trench. The hypocentral distribution highlights the shallow subduction at the northern segment, which becomes steeper and deeper to the south. The pre-earthquake stress distribution, inferred from the P and T-axes of earthquake faulting mechanisms, represents the compressional fore-arc and extensional back-arc stress regimes. Shallow NNE-SSW under- thrusting and NNW-SSE opening up of the marginal sea basin stresses were observed and this trend changes to NE-SW to N-S at intermediate depths. We collected three epochs of campaign mode GPS data along the arc from May 2002 to September 2004. These observations show nearly pure convergence along the Andaman trench prior to the earthquake. During this period the GPS sites moved westward relative to India at ~5.5 mm/yr, consistent with the earlier results. Along arc GPS velocity vectors suggest that the Andaman trench is part of a purely slip partitioned boundary, with the strike- slip component of the India-Sunda relative plate motion being taken up on the transform fault in the Andaman Sea or on the West Andaman Fault, and the convergent component on the Andaman trench. Although near normal convergence was observed, it sampled only a fraction of a possible full Andaman microplate convergence velocity, because elastic deformation from the locked shallow megathrust caused displacements toward the overriding plate, that is, away from India. Based on the Indian plate velocity and Andaman spreading rates, this component amounts to ~85% of the pre-seismic convergence. These geodetic velocities represent the present day geologic deformation rates and have been used to build an empirical relation for replenishment time required for megathrust occurrence. For Indo/Andaman convergence rates obtained from this study, renewal rates of ~630-1100 years are observed. Assuming the convergence represents the lower limit of the deformation happening there, which may actually vary over the entire seismic cycle, a value of ~1000 years may be appropriate for a 2004 type megathrust earthquake recurrence along the Andaman – Nicobar margin.
S24A-02 INVITED
Coseismic, Postseismic, and Interseismic Deformation, and Long-Term Segmentation Near the Boundary of the 2004 and 2005 Sunda Megathrust Ruptures
Simeulue Island, off the west coast of northern Sumatra, straddles the boundary of the 2004 and 2005 Sunda megathrust ruptures. The 2004 and 2005 earthquakes nucleated northwest and southeast of Simeulue, respectively, and each ruptured bilaterally toward the 100-km-long island. Cumulative uplift was 1.5 m at both the northwest and southeast tips of the island but diminished toward the island's center, where uplift was 0.5 m or less. Hence, although the 2004 and 2005 uplifts overlapped, there was an uplift deficit, or saddle, on central Simeulue. In addition to enabling observations of coseismic uplift, Porites coral microatolls enable measurements of postseismic and interseismic elevation changes. Whereas uplift in the year after May 2005 was negligible (< 10 cm) at the northwest and southeast ends of Simeulue, 30 cm of postseismic uplift occurred during that period in central Simeulue. Moreover, analysis of microatoll morphology indicates that, averaged over decades and longer, interseismic strain accumulation rates are lower in central Simeulue than at the island's ends. In the region of the Simeulue Saddle, we find evidence for up to ~20 cm of uplift at about the time of and possibly related to the M 7.3 foreshock of Nov 2002. Additionally, at one site in central Simeulue, we see evidence for multiple moderate (30-70 cm) subsidence events and multiple moderate (70 cm or less) uplift events over the past millennium. Although, in the fossil coral record, it is difficult to distinguish between single coseismic events, earthquake couplets a few years apart, and slow slip events spanning days to years, such subsidence and moderate uplift events are rare outside the Simeulue Saddle. These moderate events appear to be loosely clustered in time. Historical intensity data imply that prior to 2005 the last great earthquake caused by rupture of the megathrust under Nias island, to the south, occurred in A.D. 1861; initial U-Th dates from uplifted fossil microatolls suggest that, like the 2005 rupture, the 1861 rupture extended north to southern Simeulue, but that uplift did not extend to northern Simeulue. Additional preliminary U-Th dates from uplifted fossil microatolls suggest that earlier earthquakes, some time around A.D. 1843 and A.D. 1799, also involved rupture of Nias and southern Simeulue. Thus it appears that the ~1799, ~1843, 1861 and 2005 ruptures were similar in that they extended into but not through the Simeulue Saddle. However, at sites in southeastern Simeulue, uplifts in ~1799 and in ~1843 were much smaller than in 2005, and the ~1799, ~1843, and 1861 events may be regarded as a triplet. Altogether, these observations suggest that the Simeulue Saddle is a poorly coupled segment of the megathrust that serves as a persistent barrier to rupture.
S24A-03 INVITED
A paleotsunami record from marshlands in West Aceh Province, Indonesia
Constraining the frequency and magnitude of large events in the Indian Ocean region is critical to assess and mitigate tsunami risk along this densely populated coastline in the future. As historical records of large tsunamis in the area are sparse, the geological record provides the best evidence of recurrence rates and size of ancient tsunamis. Based on sediment data from coastal marshland deposits we present a paleotsunami record for West Aceh Province, Indonesia, an area immediately adjacent to the seismic source that was severely affected by the December 2004 tsunami. The recent tsunami deposited a distinct, typically 10-20 cm thick sand sheet up to 2 km inland within a prograding beach ridge plain. Sediment cores from swales in between beach ridges revealed three older sand layers, up to 10 cm thick, and intercalated within organic-rich marshland deposits. At least two of the older sand layers can be followed for several hundred meters along shore normal transects. Coring sites of different transects can be laterally correlated by following pronounced older beach ridges running parallel to the shoreline. The three individual sand layers occur at different distances to the shoreline, with the youngest sand layer at ~500 m distance from the present coast and the oldest one between 1500 m to 2000 m inland within older beach ridge complexes. The spatial distributions as well as grain size trends suggest landward directed flows over a prograding beach ridge plain, which can be best explained by ancient tsunamis. Radiocarbon dating of these deposits indicate three events occurring around 1000 AD, between 1350AD-1550AD, and after 1800AD, with the latter potentially correlating with a historically reported event in 1907AD.
S24A-04
Thai Evidence for Recurrent Indian Ocean Tsunamis of the Last 3,000 Years
Late Holocene tsunamis probably like the one in 2004 have attacked Thailand's Andaman Sea coast at irregular intervals of less than 1,000 years on average. The 2004 tsunami ended a recurrence interval ~600 years long. Such geology could have foretold the 2004 Indian Ocean tsunami. These interpretations, all preliminary, are based on stratigraphic reconnaissance of marshy swales between grassy beach ridges of Ko Phra Thong, an island 125 km north of Phuket. Pre-2004 tsunamis best explain several properties of sand lentils that are intercalated with freshwater peaty soils of the swales: SETTING. The island's beach-ridge plain, which extends 3 km inland and 15 km coastwise, was mostly overrun by the 2004 tsunami, which coated much of it with sand. By contrast, river floods do not reach the plain because tidal inlets separate Ko Phra Thong from the mainland. Storm surges of recent decades scarcely penetrated the beach-ridge plain. Storm runoff from the beach-ridge crests, which gently rise just 1-2 m above the swales, is limited by permeable soils; the ridges lack rills and sheetwash fans. THICKNESS. Like the 2004 tsunami deposit, three of the sand lentils maintain thicknesses of 5-20 cm along a cross-swale extent of 20-40 m, as seen in correlated pits and a trench 0.5 km inland from the present beach. ARCHITECTURE. Along this cross section and another nearby, each of the sand lentils extends farther up the seaward flank of the swale than up the landward side. The 2004 sand sheet is thick on the seaward flank and thin on the landward flank. FREQUENCY. In a sequence of three sand lentils, the youngest is close to 550-700 years old and the two earlier ones probably formed between 2,000 and 2,800 years ago. These estimates are based on radiocarbon dating of leaves and bark in buried soils and of marine shells below the lowest peat. RECYCLING. Two of the lentils contain leaf fragments centuries older than the time of deposition. The discordance is consistent with tsunami-induced scour of the beds of tidal inlets. Puzzles about the sand lentils include: MICROFOSSILS. Sand lentils examined thus far contain few calcareous foraminiferal tests and no diatom valves. Both these fossil types are abundant in Ko Phra Thong's 2004 tsunami deposits. In addition, the fossils of freshwater diatoms have been found in the peaty soils of the swales, both beneath the 2004 deposits and beneath the underlying sand lentils as well. Did tropical diagenesis erase fossil diatoms from the sand lentils? INUNDATION. Along the cross sections studied thus far, the maximum flow depth in 2004 probably exceeded 10 m. Do the sand lentils represent similar inundation? TSUNAMI SOURCES. In one of the cross sections, traces of a sand bed 1-2 cm thick are preserved between two of the sand lentils. Does the thin bed signify a tsunami source little larger than that of 1881, while the thicker sand lentils signify full-length Sumatra-Andaman ruptures?
S24A-05
Possible Earthquake Generated Turbidites along the Sumatra Margin
An international science party of 37 embarked from Phuket Thailand to collect cores from the Sumatra margin to investigate the paleoseismic record from great earthquake rupture in the Holocene. 99 piston, gravity, kasten, and multicores were collected along the length of the Sumatra margin, from the 2004 rupture zone in the north, to the southern tip of Sumatra Island. Cores were collected in the trench and in lower slope piggyback basins. The trench lithology was dominated by fine mature quartz sand, consistent with the well known Himalayan source of the accreting Bengal and Nicobar fans, and contained several tephras. Slope basins contained similar lithology, with abundant forams, and significant organic debris in those basins near the offshore islands of the Sumatran forearc. Core sites were located in places that optimized for 1) isolation from confounding factors, like terrestrial storm generated sediment input, 2) sufficient sampling for structural segment boundary determination, and 3) preservation of sediments that could permit our analytical methods. Our strategy was to densely sample both trench and basin sites to test correlations between sites to determine whether observed turbidites are earthquake generated. If so, the sampling density may allow discrimination of segment boundaries as well as event histories for margin segments. Preliminary analysis suggests that the cores contain turbidites most likely generated by the 2004 and 2005 northern Sumatra great earthquakes. These are represented by a large shallow multipulse event overlain by a smaller single pulse event at the seafloor, with no observed hemipelagic sediment between them. Ongoing 14C and Pb210 dating with stratigraphic correlation will test the origins and connectivity of these and numerous other Holocene turbidites.
S24A-06
Paleoseismoloigcal evidence of the 1762 and earlier earthquakes off Myanmar
The potential for giant tsunamigenic earthquakes off the west coast of Myanmar, north of the 2004 Sumatra- Andaman earthquake has been debated. Tectonic environments (Cummins, 2007), recent GPS observations (Socquet, 2006) and historical descriptions of geomorphological changes and eyewitness accounts of the 1762 Bengal earthquake seem to support that the earthquake potential exists. As possible evidence of past earthquakes, marine terraces have developed along the Rakhine coast. Our field surveys at the four localities on west and middle Phayonkar islands identified at least three clear steps of marine terraces approximately 1-2 m, ~5 m and ~10 m above present sea level. These steps indicate that sudden sea level change, or coseismic uplift associated with the large earthquakes, occurred in the past. Radiocarbon dating of coral or shell fossils provide ages of the uplift events. Six samples from the lower terrace indicate the most recent uplift occurred between 1585 and 1860 AD. This range includes the 1762 Bengal earthquake. Dates of the middle step, based on four samples, range between 805 and 1220 AD. Four samples on the upper step indicate an uplift event between 1295 and 600 BC. The time interval between the lower and middle step is several hundred years, while that between the middle and upper step is more than 1000 years. The longer time span and the larger elevation difference of the earlier interval may indicate either missing event(s) or variability of earthquake size and recurrence interval.
S24A-07 INVITED
The Potential for Giant Tsunamigenic Earthquakes in the Northern Bay of Bengal
The Great Sumatra-Andaman Earthquake and Indian Ocean Tsunami of 2004 came as a surprise to most of the earth science community. While it is now widely recognised that the risk of another giant earthquake is high off central Sumatra, just east of the 2004 earthquake, there seems to be relatively little concern about the subduction zone to the north, in the northern Bay of Bengal along the coast of Myanmar. It is shown here that similar indicators suggest the potential for giant earthquake activity is high: (1) the tectonic environment is similar to other subduction zones that experience giant megathrust earthquakes; (2) stress and crustal strain observations indicate the seismogenic zone is locked; and, (3) historical earthquake activity indicates that giant tsunamigenic earthquakes have occurred in the past. These are all consistent with active subduction in the Myanmar subduction zone, and it is hypothesized here that the seismogenic zone there extends beneath the Bengal Fan. The results suggest that giant earthquakes do occur off the coast of Myanmar, and that a very large and vulnerable population is thereby exposed to a significant earthquake and tsunami hazard.
S24A-08
Predecessors of the 2004 Indian Ocean Tsunami: Inferences Based on Historical, Archeological and Geological Evidence From the Indian Coast and the Andaman-Nicobar Islands
The 2004 tsunami is an unprecedented event in the Indian Ocean. Never in the recent or distant history of the region has such a transoceanic event of devastating proportion is known to have been reported. Obviously, apparent lack of historical references in the culturally ancient Southeast Asian region suggests rarity of such events. Therefore, a major question that has been posed since the 2004 tsunami is whether similar events have occurred in the region in the past. If there are predecessors, what is the frequency of such events? Resolving this question is of crucial importance in developing the recurrence history of megathrust earthquakes and assessing the tsunami hazard of the region. Our strategy has been to tackle this problem using historical and archeological data, combined with geological investigations in the affected regions of the Indian coast, including the Andaman- Nicobar Islands. Citations from south India on ancient tsunami include classic Tamil texts, which mention about a devastating sea surge around A.D. 950 in the southeastern coast of India. Our studies were focused on two ancient port cities on the east coast of India: Mammallapuram and Kaveripattinam, the latter being a major township during the first millennium. The 2004 tsunami had scoured Mammallapuram beach exposing the basements of older temples. We have identified a discordant sand deposit sandwiched between two bricklayers at a site where the ruins of different generations of temples have been excavated. The radiocarbon dates suggest that this was deposited during 955+/-30 yr B.P., close to the historically documented period of devastation of this site by a sea surge. Excavations at Kaveripattinam, located 200 km to the south, revealed a widely distributed occupation horizon of A.D. 8-10 century, marked by a superjacent sand layer. We suspect that this layer represents the A.D. 950- sea incursion mentioned in the in the classic Tamil texts, also in line with the archeological inferences made earlier from this site. The supporting geological evidence for the afore-said approximations of a 1000-year-old sea surge, possibly a tsunami, need to be found from the Andaman-Nicobar Islands, closer to the source zone. Although the database is currently preliminary, the ages of the younger terraces in the islands suggest uplift events around 900 and 2000 yr B.P. Stratigraphic evidence of the previous tsunami events should reveal either as peat-sand couplets or as drowned ancient mangrove forests. Our investigations have revealed several potential candidates, which we are exploring further. The GPS estimates (Subarya et al., 2006) suggest an average recurrence interval of 230-600 years for the 2004-type earthquakes, if all the slip is released only through similar-size events, but double that if the slip was aseismic or expended by M 8 events like those in 1881 and 1941. Our inferences agree with the latter estimate.